Investing in Artificial Intelligence Stocks and Companies
TABLE OF CONTENTS
- What is Artificial Intelligence?
- How Does Artificial Intelligence Use Big Data?
- Examples of Artificial Intelligence Use Cases
- Natural Language Processing and Speech Recognition
- Artificial Intelligence for Healthcare
- Artificial Intelligence for Computer Vision
- Global Artificial Intelligence
- Will AI Replace My Job?
- Investing in AI Stocks
Perhaps one of the most defining technologies of the century, artificial intelligence is an emerging technology with the potential to transform every company in every industry in just about every way you can imagine. That’s the promise of artificial intelligence anyways. Anyone who realized the potential of AI to generate exponential value – and consequently, exponential returns – will want to invest in artificial intelligence. Whether you’re a retiree looking to get some exposure to artificial intelligence or a CTO looking for something to plaster on the next board meeting slide deck, we’ve got you covered in this guide to investing in artificial intelligence. First, let’s start with some terminology.
What is Artificial Intelligence?
People have lots of questions about artificial intelligence because the term is used loosely to label just about anything these days. Artificial intelligence has been around since the 1960s, and it’s only recently taken off because of advanced computing hardware like GPUs from NVIDIA (more on this later).
First, we need to get some AI definitions out of the way first. You will find many of the below terms used interchangeably.
- Artificial intelligence – This is where a computer can begin to process data and infer complex relationships just like a human being can. It is the broadest term in use in that it encompasses the rest of these terms.
- Machine Learning – Machine learning is about how computers with artificial intelligence can improve over time using different algorithms (sets of rules or processes), as they are fed more big data. This process is refered to as training.
- Neural Networks – A type of machine learning, neural networks are superficially based on how the brain works.
- Deep Learning – Deep learning is simply a larger neural network. Deep learning networks typically use many layers, and often use a large number of units at each layer, to enable the recognition of extremely complex, precise patterns in data. Possible applications are endless.
- Cognitive Computing – Pretty much just a clever repackaging of AI by some marketing team. Means the same thing, depending on who you ask.
While the techies can debate among themselves the difference between “machine learning” vs. “deep learning”, we’re going to consider the two terms synonymous for the sake of simplicity. There is no one single neural network that will rule them all, but rather hundreds of algorithms that will excel in various niches.
Simply put, deep learning and machine learning involve training algorithms to do useful things by feeding them lots of delicious big data. The ability for a computer to learn more over time based on experience, something the human brain does naturally, is also referred to as cognitive computing. According to Gartner’s AI Hype Cycle for 2020, cognitive computing is now defunct, and in its place are dozens of new categories for AI. The most important one of all is artificial general intelligence or AGI.
The picks and shovels of artificial intelligence hardware are the specialized chips uses to train machine algorithms (training chips) and then the chips they run on in production (inference chips). One company that makes both is NVIDIA, a stock we first wrote about in an article titled The Artificial Intelligence Stock That Rocked Wall Street. Since then, shares of NVIDIA have skyrocketed giving them a market cap that’s almost on par with Intel, the world’s biggest chipmaker. (Intel is also making investments in AI chips.) Most of NVIDIA’s growth has come from the sale of graphical processing units (GPUs) that are used to train AI algorithms. NVIDIA isn’t the only company that makes GPUs, but they have majority market share. Aside from some temporary setbacks, NVIDIA has continued to be a leading player in AI chips – at least for now.
Chinese AI chipmakers are desperately trying to compete with NVIDIA so they don’t have to rely on the U.S. for chips. What started out as a trickle of AI chip startups has now ballooned into dozens. Across the globe, AI chip startups are developing new artificial intelligence chips. Many of these chips are highly customized and focused on inference “at the edge” doing things like always-on-voice recognition.
Over the years, the AI chip landscape has been changing. There are not many chip types being used. For example, the AI industry is now using other chip types like FPGAs for inference. Other technologies like neuromorphic computing are helping us rethink AI chip design. The entire industry is now moving towards chips that are highly customized with software that’s been optimized for specific tasks. Plenty of AI chip startups are getting funded, leaving NVIDIA investors asking, Is NVIDIA Stock Still a Good Way to Invest in AI Chips? If you’re losing sleep over that question, then check out how you can Invest in Many Types of AI Chips With One Stock. We sold some NVIDIA shares after their recent split but it’s still our biggest position and will remain that way for now until growth stalls. That’s recently happened, so now we’re talking a look at the broader semiconductor industry.
Not Everything is Artificial Intelligence
Back in the late 90s, eCommerce was going to transform the world. Some companies were adding the suffix “.com” to their business name to see their share price skyrocket as a result. Today, all you need to do is plaster the term “machine learning” all over your pitch deck to increase your valuation by multiples. Back in 2017, we issued A Warning About “Artificial Intelligence Stocks” which cautioned against stuff like this:
When AI started to make headlines on the nightly news, people started looking for AI stocks to invest in, so the pundits had to start choosing some stocks to call “AI stocks.” We wrote about how Investing in Artificial Intelligence (AI) Stocks is BS if you’re simply telling everyone to buy the FAANG stocks all over again. Even stocks that traded on major exchanges were trying to cash in on the AI hype train (Veritone seemed to be one such company, though they’ve more recently been demonstrating traction for their AI offering). Tribune Publishing claimed they were going to use AI to produce thousands of videos, and the stock soared on the news. All of these stocks have since come back down to earth making everyone realize that not everything is artificial intelligence. And even if you are using artificial intelligence, your company can still underperform the broader stock market.
Artificial Intelligence and Transparency
<Begin Rant> Nothing annoys us more than the kangaroo courts being drummed up by a small handful of armchair CEOs on Twitter. These are the same people who spend their free time thinking of ways they can find Halloween costumes offensive, and they’ve now decided that artificial intelligence tells the truth too much, and should therefore be silenced. As they often do, they’ll try to make people kowtow to their demands under the guise of representing some class of people who they have no right whatsoever ever to represent. You’ll hear them talk about “AI and transparency” while suggesting some sort of oversight committee that makes sure the AI is playing fair and treating everyone the same. This is an extremely dangerous path to go down, and we wanted to make sure to denounce this rubbish for what it is. Always look at what’s motivating people who call for AI transparency. <End Rant>
What will influence outcomes for AI algorithms is the data you feed them. What all AI algorithms have in common is that they’re only as good as the big data you’re feeding them.
How Does Artificial Intelligence Use Big Data?
We come across a lot of companies claiming to use AI to achieve some competitive advantage, and we always ask the same question. What proprietary data sets are you using to train your algorithms with such that you’ll be able to realize a competitive advantage? Truth is, machine learning algorithms are only as good as the data you train them with, a process that’s known as supervised learning. One of the first companies to start wielding the power of big data was Palantir, a secretive firm that took Silicon Valley by storm when they started to acquire startups left and right. Simply put, Palantir uses big data to solve the most important problems for the world’s most important institutions. (Palantir is now publicly traded, something we talked about in our piece on An Enterprise AI Showdown – C3 Stock vs. Palantir Stock.) Many followed in their footsteps, and by 2016, there were at least 14 billion-dollar big data startups working on harnessing the world’s big data. Some companies specialize in analyzing data exhaust from inside corporations, helping them identify threats or quickly isolate problems. Others like Zoom plan to use all the big data from unified communications to build productivity tools for knowledge workers.
Big Data vs. Data Warehouses
Big Data vs. Data Warehouses. What’s the Difference? That was the title of an article we wrote which talked about how big data lets us find “the unknown unknowns,” things we didn’t even think of to start measuring. We also need to expand our definition of data to include things like CCTV camera video feeds, audio clips, imagery, sensor outputs, temperature, and just about anything you can think of that can be measured. Smartphones are now producing geospatial data so we can learn more about how people behave. Companies like Cloudera (CLDR) are building multi-billion-dollar businesses around managing all this big data alongside many other data-focused companies, like Snowflake, which Warren Buffett invested in, data-to-everything company Splunk, or Databricks and their Data Lakehouses. (Here’s our latest pieces on Splunk and Snowflake.) Legacy data tool providers like Informatica should be avoided as they missed the boat while firms like Alteryx fill the gap. Data center REITs provide another way to play the growth of big data, but we think the risks outweigh the rewards. There’s also data storage hardware to consider as a pick-and-shovel play on big data, and Pure Storage probably leads the pack in that niche. MongoDB offers a way to play the growth of unstructured data, but unfortunately they don’t provide enough information for investors.
Crowdsourcing platforms like Appen cleanse the data, then new techniques like topological big data analysis are used to find patterns in the data. (Although Appen is publicly traded, we’re choosing not to invest in it.) Even with all the data we have, we still need more. A fair number of startups are now creating synthetic data that’s used for applications like autonomous driving. Singaporean startup Near has assembled the largest database of human behavior, which is a good segue into our next topic.
Unstructured Data and Data Privacy
Data can be divided into two types – structured data and unstructured data. Somewhere around 90% of all data is unstructured data, a catch-all term that refers to information that isn’t organized in a way that makes it easy to analyze or understand. (Sometimes this is referred to as “data exhaust.”)
All that crap that gets posted on social media is being analyzed and used for something. This very article is a piece of unstructured data that some machine learning algorithm is probably munching on right now. Everything that gets publicly posted on the Internet, anywhere, is unstructured data that’s probably being used right now by AI algorithms to glean some insights from. Since all the data is being monitored, there are some real concerns about big data privacy. And there should be. It’s something we talked about in a piece titled Big Data: Big Brother or Bigger Than Any of Us? If AI checks your grammar, assigns you a credit score, and performs your background checks, it probably knows much more about you than you think.
Enough of the sinister conspiracy theories. Let’s get back to talking about the potential of artificial intelligence to transform the world in a good way, the kind of way that would make the ESG-types rub their hands together with glee.
Examples of Artificial Intelligence Use Cases
When DARPA starts investing in AI startups, you know there’s some “there there” as Gertrude Stein would say. Venture capital firms (VCs) are pouring money into AI, and hype has reached a fever pitch which seems to resemble the dot-com bubble. There are even VCs using AI algorithms to pick winning startups to invest in. They kind of need to considering how many AI startups there are now. Last we counted, there were over 3,500 startups using artificial intelligence to do just about anything you can think of.
From preventing boat collisions to detecting counterfeit goods, the applications for machine learning are truly endless. Automated machine learning is also a thing as companies work on making artificial intelligence easy to use so that you don’t need to be a rockstar programmer to make things happen. Search engines are using AI to organize the world’s information, and researchers are using AI to expedite research. Coca-cola uses AI to sell more drinks and McDonald’s uses AI to sell more fries. We’re even using AI to help solve the world’s energy problems. It’s all very exciting.
Many artificial intelligence startups are seeing success by targeting large industry verticals:
- AI in Law Enforcement – Big data and CCTV video are being used to reduce human injuries and death in areas like road safety or public safety.
- AI in Law – Having some AI bots scouring the world’s legal information seems like a must for the modern legal office. They’re even working on chatbot lawyers.
- AI in Investment Banking – All facets of an investment bank are affected by AI, from back-office support to algorithmic trading to issuing debt. Even their clients use AI – hedge funds use AI for trading and asset managers use AI to create smart beta ETFs.
- AI in Insurance – It’s easy to imagine how predictive analytics can be used in place of human actuaries.
- AI in Real Estate – Zillow uses AI to produce more accurate estimates. AI-powered appraisals of commercial real estate are now more accurate than human estimates.
- AI in Travel – There are apps that predict airfare, and they’re even putting machine learning algorithms in control towers.
- AI in Accounting – Bean counting is right up machine learning’s alley. Even small business accounting programs are using AI.
- AI in Agriculture – From indoor farming to agricultural intelligence, AI is helping us feed more hungry mouths.
- AI in Education – AI is great for adaptive learning, where we teach people differently based on how they learn. In China, they use AI tutors to learn English.
Other startups are also working to provide solutions across industry verticals, something they’re calling “enterprise AI” or “AI as a Service.” Enterprise AI startups are getting funded left and right. Startups like Sentient and Vicarious are developing AI algorithms that can be used in any industry or for any robotics application. Many applications for AI like global identity verification, robotic process automation (RPA), and predictive analytics are industry agnostic.
AI in Sales and Marketing
Sales and marketing are part of just about every business these days. Whether you’re selling online or from a brick-and-mortar outlet, there are many ways AI can help retail stores increase sales and decrease costs. Technologies like automated checkout systems can help reduce labor costs. Behind the scenes, startups like Blue Yonder are building machine learning algorithms for inventory management and helping you figure out your pricing models. Does your customer service center in Manila need a makeover? Billion-dollar startup Afiniti uses machine learning to match customers with company agents for the best results. Customer relationship management is also being transformed by AI algorithms that can also help your enterprise sales team sell more efficiently. Now, sales managers are able to extract insights from communications between the sales team and customers, something that’s referred to as “revenue intelligence.” If your product is digital, AI can help you build better software and collect better customer market research. If your product is something people experience, AI can help read your customer’s emotions, something that’s also referred to as affective computing. If you want to sell your product online, AI-powered programmatic digital advertising can help, while AI-powered experience management can help delight your customers.
AI and Human Resources
(Warning: Rant to follow for the next paragraph.) We’re not big fans of human resources (HR) because they largely get in the way while trying to create more work for everyone. For example, if you’ve served time in a multinational corporation, you would be familiar with HR asking you to perform some mandatory function using some of the poorest engineered software known to man. The only logical conclusion we could reach is that the one-eyed wench leading the blind will sign just about any purchase order from any vendor with an app. Ever try using Concur? What about Workday in the early days? It actually took anonymous feedback people submitted to HR about their managers and sent it to the managers! True story. Anyways, human resources will buy just about anything, so we really need to vet AI HR solutions.
AI for Recruiting
In a piece titled “The Death of Recruiting by AI – It’s About Time,” we speculated that competent tech startups will start to deliver effective AI solutions, and all these “boutique” recruitment firms (where the only qualification to start one is that you can fog a mirror) will start to lose out on all those exorbitant recruiting fees they were raking in to find tech people. For example, ZipRecruiter’s AI-powered algorithm learns what each employer is looking for and provides a “personalized, curated set of highly relevant candidates” based on more than 60 factors. They’re now a unicorn. We’re also using AI algorithms to identify talented software engineers and people who are less likely to bolt after they get hired. Inside the workplace, we’re using AI algorithms to monitor employee Internet usage or help figure out which types of people will play well together using psychometrics.
(Warning: Rant continues.) Startups that use AI for recruiting are popping up left and right, and people need to understand that there is a right way and a wrong way to use AI for recruiting. The wrong way is trying to use AI to achieve equality of outcome. Regardless of where you stand politically, attempting to use recruitment technology for promoting equality of outcome is a dangerous approach that does a disservice to every individual who wants to be evaluated on their own merits instead of being patronized based on some attribute that is irrelevant to their ability to perform.
You will always have a spectrum of competence. You must pay the upper bound higher than the lower bound to retain talent. Gender should never come into play. Neither should the color of someone’s eyes, their height, their weight, their food preferences, their sexual preferences, or any other attribute except what counts – competence. Any business that proposes we change this model and start rewarding people based on anything but competence needs to be condemned from the highest pulpit.
Anyways, we’ve gotten a bit off track but the point here is that there are certain applications that AI just won’t work well for. Cybersecurity isn’t one of them.
AI for Cybersecurity
In our guide to Investing in Computing Stocks and Companies, we talked about how cybersecurity is a high-growth sector which investors are pouring money into. That’s because hackers – 90% of whom are financially motivated – are becoming increasingly sophisticated in their attacks. Maybe it’s time for the machines to start defending themselves for a change. Back in 2017, we saw AI cybersecurity startups start to pop up. Now, there are so many getting funded we can barely keep track of them all. There are even some AI cybersecurity stocks for retail investors to get in on the action.
Most anti-virus software relies on a library of known types of malware and other types of malicious software. This means there is a need to push constant updates to customers as new threats emerge. Contrast this approach to cybersecurity startups like Crowdstrike which use algorithms to spot suspicious activity that hasn’t been previously identified. Another startup called Darktrace uses machine learning algorithms to sniff out every single device on your network so that it can create a virtual immune system to detect cyberthreats. As for passwords, those are going away. Startups like BioCatch are developing something called “behavioral biometrics,” which collects and analyzes over 500 traits to make sure it’s you. These include hand-eye coordination, pressure, hand tremors, navigation, scrolling, how you type your name, and other finger movements, such as how you wake up your computer when it falls asleep.
AI Algorithms Get Artistic
Machine learning algorithms are also being used for more creative applications like designing logos, writing clever headlines, or aggregating your news feed. At least 10 startups claim to be using AI in the world of fashion and at least 11 startups are composing music with AI. Even your fashion stylist over at Stitch Fix is an AI algorithm. They’re even using AI algorithms to herd cattle. Of course, not all AI applications will work.
AI For Everything
Since artificial intelligence is becoming so pervasive, it’s easy to say that every company uses artificial intelligence. If your company throws up a chatbot that uses natural language processing, you may be using AI, but so is everyone else. Many popular artificial intelligence software frameworks are free. Again, it’s all about the big data.
We visited with Merantix, an AI incubator in Berlin, and their founder taught us something important. An AI startup needs to add exponential value in order to create exponential returns. Merantix has somewhere around 700 ideas for artificial intelligence applications, and they want to distill that down to the ones that will be worth billions. We decided to deep dive into a few domains where artificial intelligence is excelling right now at solving big problems – healthcare, computer vision, and natural language understanding.
Natural Language Processing and Speech Recognition
The term natural language processing (NLP) refers to the methods which AI algorithms use to understand human speech in written, verbal, and non-verbal forms. Those who belong to the Apple cult are familiar with one popular implementation of natural language processing – Siri. (Interesting side note. The co-founder of Siri, Dag Kittlaus, built another NLP startup called Viv which he sold to Samsung.) The category of natural language processing includes everything from algorithms that write news articles to interpreting sign language for the deaf using computer vision. Speech recognition, chatbots, transcriptions, translations, it’s all in our guide to Investing in Natural Language Processing.
Artificial Intelligence for Healthcare
The industry which is being transformed the most by artificial intelligence right now is healthcare. That’s why we’ve broken this out into a separate guide on what areas of the healthcare industry are being transformed by AI. (Hint: It’s pretty much all of them.) Over the past few years, we spent time speaking with researchers at the hallowed campus of Stanford and visiting with founders in the Silicon Valley Vatican to learn how areas like medical imaging and drug discovery are leveraging AI to do things that would blow your mind. The explosion of data in the healthcare industry is enabling predictive analytics startups to offer personalized medicine to the masses. It’s all covered in our guide to Investing in AI Healthcare Companies and Stocks.
Artificial Intelligence for Computer Vision
Computer vision might sound like the latest 3D eyewear, but it’s actually a field of research for designing machines with the ability to process, understand, and use visual data just as humans do. Large Chinese startups have been dominating computer vision with applications such as facial recognition, while other countries look on nervously, afraid of the power it wields. For investors, much of the opportunity lies in areas such as factory automation. So much is going on in this space that we created an entire guide around it – Investing in Computer Vision Companies and Stocks.
Global Artificial Intelligence
The only thing we love more than technology is traveling the world to meet people smarter than us that are building great companies. Across the globe, there’s a global AI race taking place where every country is competing to build the best AI algorithms. In the words of Vladimir Putin, whoever leads in artificial intelligence will rule the world. China and the United States are in the lead presently, but it’s anyone’s race. You can read all about it and more in our 6,400-word guide on The Global AI Race – Top AI Startups Around the World.
Will AI Replace My Job?
Short answer, probably. Whether you’re blue collar or white collar, your job could be replaced at any time. The better question to ask is if you’re actively working to make your role irreplaceable? If you’re not, then it doesn’t matter if John-in-Mumbai steals it or an AI algorithm does. We also need to think about all the new jobs being created by the emergence of AI as well. If mobility is important, there are plenty of human-in-the-loop type jobs you can work from anywhere. You won’t become an AI trillionaire, but you’ll get to send your mates pics of your over-priced Mercedes van parked in front of idyllic settings. Do it for the gram, bruv.
We wrote about nine of the best jobs for the future and a running theme was AI, robotics, and big data. The machines aren’t going to train themselves, at least for now. The best way to learn artificial intelligence is to first decide what you wish to accomplish. There are many AI courses and certifications that can make you relevant again. Then, you can use AI to replace other people’s jobs with robotic process automation – at least until the machines learn to code themselves.
Robotic Process Automation or RPA
Simply put, robotic process automation (RPA) is when we teach machine learning algorithms how to do white-collar jobs that have a certain element of predictability to them. Think data entry automation or automated client service desks. You may be tempted to think that robotic process automation (RPA) is just marketing speak. After all, we’ve been working on replacing humans with software since the late 90s when Rational was replacing software testing humans with “software robots”. The difference between automating tasks with software and robotic process automation is that the latter uses AI algorithms to learn how to do a job and – this the important part – adapt to changes dynamically. Not having to engage an expensive developer every time something changes is a huge benefit of using RPA. It’s about handling exceptions as they come along instead of embedding them in code.
RPA has taken off like wildfire now. Leading RPA startup Automation Anywhere is now a $6.8 billion company having taken in $840 million in funding. Another leader in this space is UiPath, a $10 billion company that’s now moving beyond RPA into hyperautomation. Some startups are targeting specific industries, like AI startup Olive which is focused on RPA for healthcare. There are even some publicly traded RPA stocks, such as UiPath, which is why we put together a piece on The “Best” Robotic Process Automation (RPA) Stocks. Lately, UiPath has been taking a drubbing along with all tech stocks but it’s never looked better. It’s also a great segue into our next section.
Investing in AI Stocks
Before we tuck in here, please make sure you read the section in this guide called “Not Everything is Artificial Intelligence.” Just because a company uses a few chatbots, it doesn’t mean they’re “an AI company.” The best question to ask a company that claims they’re “doing AI” is what proprietary datasets do they have to train their algorithms that nobody else has? We always pose this question to startups we meet with. When it comes to assessing stocks, it’s equally as tough. That’s why we wrote this guide.
Also, be wary of anyone who tries to peddle you some method of using machine learning for stock trading strategies. As we said before, hedge funds have been all over this for decades. Anyone who claims to be selling a tool to retail investors that uses artificial intelligence to generate alpha is full of it. Thinking you can day trade your way into wealth using any strategy being peddled – AI or not – is a recipe for disaster. It’s something we cover in our Complete Guide to Buying Stocks for Beginners.
As recent as 2016, retail investors couldn’t invest in artificial intelligence stocks because there weren’t any except for perhaps NVIDIA. Fast forward to today and every talking head out there has a different opinion about what constitutes an AI stock. (There is no industry classification for artificial intelligence.) We spent the last six years separating the wheat from the chaff. The FAANG tech stocks are not AI stocks, and no penny stock on the face of this earth will ever make you financially independent.
If you are willing to look outside your backyard, you’ll find some very interesting pure-play stocks across the pond. In addition to the AI stocks we’ve already talked about so far today, we put out a few more guides for AI investors on Investing in AI Healthcare Stocks and Investing in Computer Vision Stocks. We also put together a general guide on How to Invest in AI stocks for retail investors. If you read through all that content and still want more, sign up as a paid subscriber and we’ll also send you the below reports.
7 pure-play AI Healthcare stocks
Comprehensive company profiles
Insightful analysis and financial analytics
What investors should watch for
5 pure-play Machine Vision stocks
Comprehensive company profiles
Insightful analysis and financial analytics
What investors should watch for
There’s also one other thing that investors need to think about here.
Artificial intelligence is the new oil. It is pervasive and being used by nearly every major company now in some capacity. Investors need to ask themselves, are there companies in my portfolio that are not adopting artificial intelligence? Those that don’t won’t be able to compete with those that do. As dividend growth investors, we love to see companies like Walmart using AI to create efficiencies that will fund dividend growth for many years to come. Any brick-and-mortar store that doesn’t start looking at retail automation to create efficiencies won’t be around for very long.
Lastly, we shouldn’t have to say this, but we’re going to anyways because some of you will still ask us about some penny stock they’re thinking of using to empty their savings account into. Do not even consider investing in any penny stocks or over-the-counter stocks no matter what drivel management spews out in glossy investors decks and flashy websites. If you want to lose money that badly, just burn your money instead. At least then it will give you some fleeting warmth.
Tech investing is extremely risky. Minimize your risk with our stock research, investment tools, and portfolios, and find out which tech stocks you should avoid. Become a Nanalyze Premium member and find out today!